Disclosure of Invention
The invention aims to provide a high-efficiency time domain broadband beam forming circuit and method adopting a cascade FIR transverse filtering structure.
The technical solution for realizing the purpose of the invention is as follows: a high-efficiency time domain broadband beam forming circuit adopting a cascaded FIR transverse filtering structure comprises N channel delay and equalization modules, 1 data synthesis module and 1 digital low-pass filtering and channel equalizer coefficient calculation module, wherein the N channel delay and equalization modules are all connected with the data synthesis module and transmit processed data to the data synthesis module, each channel delay and equalization module comprises a variable integer/fractional delay module, a digital down-conversion module and a digital low-pass filtering and channel equalization integrated module which are sequentially connected, the variable integer/fractional delay module transmits delayed data to the digital down-conversion module, the digital down-conversion module transmits orthogonal baseband data to the digital low-pass filtering and channel equalization integrated module, and the down-conversion digital module also transmits the data to the digital low-pass filtering and channel equalizer coefficient calculation module, and the digital low-pass filtering and channel equalizer coefficient calculating module transmits the calculated filter coefficient to the digital low-pass filtering and channel equalization integrated module.
A working method based on the circuit comprises three working modes, namely: the method comprises an initialization mode, a channel equalization mode and a normal working mode, wherein the initialization mode comprises the following steps:
step 1, determining a coefficient of a Farrow filter according to a set delay precision and a delay error in a bandwidth, and putting a fixed coefficient of a variable fraction delay Farrow filter of a variable integer/fraction delay module;
step 2, setting the variable FIR filter coefficient of the digital low-pass filtering and channel equalization integrated module as the coefficient of a P-order digital low-pass filter;
step 3, setting the integral multiple of the clock period and the fractional multiple delay of the variable integer/fractional delay module to 0;
the channel equalization mode comprises the following steps:
a, a digital low-pass filtering and channel equalizer coefficient calculating module calculates the equalizer amplitude-phase characteristics of each channel by adopting a frequency domain channel equalization method according to the output data of each channel digital down-conversion module;
b, multiplying the amplitude-phase characteristic of each channel equalizer obtained in the step A by the amplitude-phase characteristic of an ideal digital low-pass filter by a digital low-pass filtering and channel equalizer coefficient calculating module to obtain the combined amplitude-phase characteristic of each channel, and respectively optimizing the coefficient of the P-order digital low-pass filter of each channel to approximate to the respective combined amplitude-phase characteristic;
setting the coefficient of a variable coefficient FIR filter of each channel digital low-pass filtering and channel equalization integrated module as a corresponding P-order optimization coefficient;
the normal operation mode comprises the following execution steps:
step a, calculating delay values of all channels according to the requirement of the change of the beam pointing direction, converting the delay values into integral multiples of clock periods and fractional time delays, and adjusting in real time;
b, setting the integral multiple of the clock period and the fractional multiple delay of each channel variable integer/fractional delay module according to the result;
c, each module of the whole circuit works by adopting a parallel pipeline structure, namely, data is delayed for a corresponding clock period through a variable integer/fraction delay module; the delayed data enters a digital down-conversion module again to become I/Q data of baseband quadrature; the I/Q data with orthogonal baseband is processed by a digital low-pass filtering and channel equalization integrated module to obtain broadband beam forming data with low data rate; the broadband beam forming data enters a data synthesis module and a beam is output.
Compared with the prior art, the invention has the following advantages: 1) the high-efficiency time domain broadband beam forming circuit adopting the cascade FIR transverse filtering structure can realize real-time broadband digital beam forming, effectively solve the problem of broadband beam directional diagram distortion caused by aperture effect and compensate mismatch among channels. 2) The digital processing circuit suitable for engineering realization adopts a cascade FIR transverse filtering structure to realize high-efficiency time domain broadband beam forming and broadband equalization among channels, and effectively reduces resource consumption by adopting a delay function preposition and FIR filter integrated design optimization technology. 3) The invention can be widely applied to radar digital array antenna, communication intelligent antenna, microphone array, sonar array and other systems.
Detailed Description
The invention provides a digital time domain delay and digital channel balanced broadband beam forming circuit for solving the problems of aperture effect and channel mismatch in a broadband digital array radar, realizes accurate beam pointing and compensates mismatch between channels. The circuit adopts a cascaded FIR transverse filtering structure to realize efficient time domain broadband beam forming and broadband balance among channels, and effectively reduces resource consumption through the integrated design of partial circuits.
The invention adopts a high-efficiency time domain broadband beam forming circuit with a cascade FIR transverse filtering structure, which consists of N channel delay and equalization modules, 1 data synthesis module and 1 digital low-pass filtering and channel equalizer coefficient calculation module, wherein each channel delay and equalization module consists of a variable integer/fractional delay module, a digital down-conversion module and a digital low-pass filtering and channel equalization integrated module. The integral multiple time delay of the variable clock period of the variable integer/fraction time delay module is realized by an FIFO (first in first out) structure, and the fractional time delay of the variable clock period adopts a Farrow filter; the digital down-conversion module adopts an orthogonal digital down-conversion method to periodically take the original value, zero and negate the original value of the original data; the digital low-pass filtering and channel equalization integrated module adopts a variable coefficient FIR filter to realize the functions of low-pass filtering and channel equalization.
The invention discloses a working method of a high-efficiency time domain broadband beam forming circuit adopting a cascade FIR transverse filtering structure, which comprises three working modes, namely:
1. an initialization mode: setting a fixed coefficient of a variable fractional delay Farrow filter of the variable integer/fractional delay module, setting a coefficient of a variable coefficient FIR filter of the digital low-pass filtering and channel equalization integrated module as a coefficient of a P-order digital low-pass filter, and simultaneously setting the integral multiple of the clock period and the fractional time delay of the variable integer/fractional delay module as 0.
2. Channel equalization mode: the digital low-pass filtering and channel equalizer coefficient calculating module calculates the equalizer amplitude-phase characteristics of each channel, the equalizer amplitude-phase characteristics and the equalizer amplitude-phase characteristics of the digital low-pass filters are synthesized to obtain combined amplitude-phase characteristics, the P-order digital low-pass filter coefficients of each channel are respectively optimized to approximate the respective combined amplitude-phase characteristics, and the optimized coefficients are arranged in the variable coefficient FIR filter of the digital low-pass filtering and channel equalization integrated module of each channel.
3. And (3) a normal working mode: setting the integral multiple of clock period and fractional time delay of each channel variable integer/fractional time delay module calculated according to the requirement of the change of the beam pointing direction, wherein each module of the whole circuit works by adopting a parallel pipeline structure, namely, data firstly passes through the variable integer/fractional time delay module to delay corresponding clock period, then enters the digital down-conversion module to be changed into I/Q data with orthogonal base bands, then passes through the digital low-pass filtering and channel balancing integrated module to obtain broadband beam forming data with low data rate, and finally enters the data synthesis module to output beams.
As described in more detail below.
A high-efficiency time domain broadband beam forming circuit adopting a cascaded FIR transverse filtering structure comprises N channel delay and equalization modules, 1 data synthesis module and 1 digital low-pass filtering and channel equalizer coefficient calculation module, wherein the N channel delay and equalization modules are all connected with the data synthesis module and transmit processed data to the data synthesis module, each channel delay and equalization module comprises a variable integer/fractional delay module, a digital down-conversion module and a digital low-pass filtering and channel equalization integrated module which are sequentially connected, the variable integer/fractional delay module transmits delayed data to the digital down-conversion module, the digital down-conversion module transmits orthogonal baseband data to the digital low-pass filtering and channel equalization integrated module, and the down-conversion digital module also transmits the data to the digital low-pass filtering and channel equalizer coefficient calculation module, and the digital low-pass filtering and channel equalizer coefficient calculating module transmits the calculated filter coefficient to the digital low-pass filtering and channel equalization integrated module.
The variable integer/fraction delay module comprises an integer delay unit and a fraction delay unit which are sequentially connected, wherein the integer delay unit adopts an FIFO structure to realize variable clock cycle integral multiple delay, and the fraction delay unit adopts a Farrow filter to realize variable clock cycle fractional multiple delay.
The digital low-pass filtering and channel equalization integrated module adopts a variable coefficient FIR filter and is used for realizing low-pass filtering extraction and channel equalization.
Preferably, the variable integer/fractional delay module is configured to write data of each channel into a respective FIFO for buffering, and implement variable integer multiple delay of clock cycle by controlling the clock cycle delay amount of data read at the output end of the FIFO structure; the module optimizes the Farrow filter structure and the related coefficient thereof according to the delay precision and the delay error requirement in the bandwidth, and realizes variable clock period fractional time delay by adjusting the fractional delay parameter of the Farrow filter in the using process.
The digital down-conversion module is realized by the following modes: when sampling frequency fsSum signal center frequency f0Satisfy fs=4f0Or 3fs=4f0In the process, the orthogonal down-conversion local oscillation sequence only needs to periodically select three values of-1, 0 and 1, namely, the original data is periodically subjected to operations of taking the original value, taking zero and taking the inverse, so that the digital down-conversion function can be realized.
Preferably, the implementation scheme of the digital low-pass filtering and channel equalization integrated module is as follows: and performing joint optimization by using a digital low-pass filtering and channel equalizer coefficient calculation module according to the synthetic amplitude-phase characteristics of the P-order digital low-pass filter and the equalizer to obtain a filter coefficient, and realizing the functions of low-pass filtering and channel equalization by using a P-order variable coefficient FIR filter.
A working method based on the circuit comprises three working modes, namely: the method comprises an initialization mode, a channel equalization mode and a normal working mode, wherein the initialization mode comprises the following steps:
step 1, determining a coefficient of a Farrow filter according to a set delay precision and a delay error in a bandwidth, and putting a fixed coefficient of a variable fraction delay Farrow filter of a variable integer/fraction delay module;
step 2, setting the variable FIR filter coefficient of the digital low-pass filtering and channel equalization integrated module as the coefficient of a P-order digital low-pass filter;
step 3, setting the integral multiple of the clock period and the fractional multiple delay of the variable integer/fractional delay module to 0;
the channel equalization mode comprises the following steps:
a, a digital low-pass filtering and channel equalizer coefficient calculating module calculates the equalizer amplitude-phase characteristics of each channel by adopting a frequency domain channel equalization method according to the output data of each channel digital down-conversion module;
b, multiplying the amplitude-phase characteristic of each channel equalizer obtained in the step A by the amplitude-phase characteristic of an ideal digital low-pass filter by a digital low-pass filtering and channel equalizer coefficient calculating module to obtain the combined amplitude-phase characteristic of each channel, and respectively optimizing the coefficient of the P-order digital low-pass filter of each channel to approximate to the respective combined amplitude-phase characteristic;
setting the coefficient of a variable coefficient FIR filter of each channel digital low-pass filtering and channel equalization integrated module as a corresponding P-order optimization coefficient;
the normal operation mode comprises the following execution steps:
step a, calculating delay values of all channels according to the requirement of the change of the beam pointing direction, converting the delay values into integral multiples of clock periods and fractional time delays, and adjusting in real time; the formula for calculating the delay value of each channel according to the requirement of the change of the beam pointing direction is as follows:
τn=nd sinθ0/c n=1,2,...,N
wherein d represents the array element spacing, θ0Denotes the beam pointing direction, c denotes the speed of the electromagnetic wave propagating in the vacuum, N denotes the number of channels, τnIndicating the delay value of the nth lane.
B, setting the integral multiple of the clock period and the fractional multiple delay of each channel variable integer/fractional delay module according to the result;
c, each module of the whole circuit works by adopting a parallel pipeline structure, namely, data is delayed for a corresponding clock period through a variable integer/fraction delay module; the delayed data enters a digital down-conversion module again to become I/Q data of baseband quadrature; the I/Q data with orthogonal baseband is processed by a digital low-pass filtering and channel equalization integrated module to obtain broadband beam forming data with low data rate; the broadband beam forming data enters a data synthesis module and a beam is output.
The high-efficiency time domain broadband beam forming circuit adopting the cascade FIR transverse filtering structure can realize real-time broadband digital beam forming, effectively solve the problem of broadband beam directional diagram distortion caused by aperture effect and compensate mismatch among channels.
The present invention will be described in further detail with reference to examples.
Example 1
The invention provides a high-efficiency time domain broadband wave beam forming digital processing circuit adopting a cascaded FIR transverse filtering structure aiming at the problems of inaccurate wave beam pointing and mismatch among channels caused by an aperture effect, thereby realizing the broadband wave beam forming with accurate wave beam pointing and compensating the mismatch among the channels. Fig. 1 shows a general block diagram of a digital processing circuit, and fig. 2 shows a detailed block diagram of each module, all logic functions of the circuit can be implemented in a single-chip FPGA, and mainly implement functions of digital time domain delay, digital down conversion, filtering extraction, channel equalization, beam forming, and the like. The circuit consists of N channel delay and equalization modules 110, 1 data synthesis module 120, and 1 digital low pass filter and channel equalizer coefficient calculation module 130. Each channel delay and equalization module 110 is composed of a variable integer/fractional delay module 111, a digital down-conversion module 112, and a digital low-pass filtering and channel equalization integrated module 113.
The structure of a commonly used broadband digital beam forming and channel equalizing circuit is that data is converted into two paths of baseband orthogonal I/Q data after down-conversion and filtering extraction, and then the mismatch between channels is equalized through a complex equalizer. On the basis, the equalized data can be processed by a variable integer/fractional delay module to complete the broadband beam forming.
However, the above structure has the following two problems: 1) after digital down conversion, the I/Q two paths of data need to be filtered and extracted, each path of each path needs to use a digital low-pass FIR filter, and the nature of the equalizer is also the FIR filter, although the functions are different, the structure is repeated; 2) time-domain delay broadband beam forming is performed after equalization, which means that two I/Q branches of each channel need to be connected with a variable integer/fractional delay module, and resource consumption is large.
In order to efficiently realize the time delay broadband beam forming and the channel equalization, the invention corrects the structure. 1) The digital low-pass FIR filter and the equalizer are subjected to synthesis optimization design and are realized by adopting a variable coefficient FIR filter. 2) The variable integer/fractional delay module is adjusted before quadrature down-conversion and digital filtering decimation. The delay amount of the variable integer/fractional delay module can be controlled, so that the delay amount of the variable integer/fractional delay module of each channel can be set to be 0 when channel equalization is carried out, and the mismatch among the channels is only inherent inconsistency among the channels. The number of the variable integer/fractional delay modules of each channel is reduced from two to one, so that the use of resources is reduced.
Variable integer/fractional delay module 111: the processing method for the variable clock period integral multiple time delay is to write the data of each channel into the FIFO structure 1111 for caching, and is realized by controlling the clock period time delay of the data read at the output end of the FIFO structure 1111; the variable fractional delay of the clock cycle is realized by a variable fractional delay Farrow filter 1112, and the Farrow filter 1112 is optimized according to the set delay precision and the delay error requirement in the bandwidth, and is realized by adjusting the fractional delay parameter.
Digital down-conversion module 112: digital down-conversion is achieved using a quadrature digital down-conversion, DDC, method 1121. The calculation formulas of the sequence values of the orthogonal digital down-conversion local oscillation are respectively
When sampling frequency fsSum signal center frequency f0Satisfy fs=4f0Or 3fs=4f0In the process, the orthogonal digital down-conversion local oscillation sequence formulas (1) and (2) only need to periodically select three values of-1, 0 and 1, so that the processing process of the orthogonal digital down-conversion is simplified into the operation of periodically selecting an original value, zero and negation for original data.
The digital low-pass filtering and channel equalization integrated module 113: to reduce the data rate, the digital down-conversion is followed by a filtering decimation. Since an equalizer needs to be inserted into each channel to compensate for mismatch between the channels, the equalizer is combined with a digital low-pass FIR filter of P order after digital down-conversion, and the combination is implemented by using a FIR filter of P order variable coefficients, and the coefficients of the FIR filter are initialized to the coefficients of the digital low-pass FIR filter. And when the channel is equalized, the coefficient of the channel is the optimized P-order digital low-pass FIR filter coefficient.
The data synthesis module 120: this module completes the summation operation of the wideband beamforming data for each channel to obtain the beam output 104.
Digital low pass filtering and channel equalizer coefficient calculation module 130: the module converts the amplitude-phase characteristic H of each channel equalizereqAmplitude-phase characteristic H of (omega) and digital low-pass FIR filterFIRAnd (omega) obtaining the combined amplitude-phase characteristics of each channel according to the formula (3), and optimizing the P-order digital low-pass FIR filter coefficient of each channel to enable the coefficient to approach the respective combined amplitude-phase characteristics.
H(ω)=Heq(ω)HFIR(ω) (3)
The digital processing circuit suitable for engineering realization adopts a cascade FIR transverse filtering structure to realize high-efficiency time domain broadband beam forming and broadband equalization among channels, and effectively reduces resource consumption by adopting a delay function preposition and FIR filter integrated design optimization technology.
Example 2
Sampling frequency fs200MHz, the echo signal is a broadband swept frequency signal with a center frequency f050MHz, and 30MHz bandwidth B. The working method of the invention comprises an initialization mode, a channel equalization mode and a normal working mode.
The steps of initializing the mode are:
step 1, calculating the coefficient of the Farrow filter (available through literature Weighted-least-square design of variable fractional-delay using coefficient symmetry) according to the requirements of the set delay precision, the delay error in bandwidth, and the like, and then putting the coefficient into the fixed coefficient of the variable fractional delay Farrow filter 1112 of the variable integer/fractional delay module 111;
according to the central frequency and the sampling frequency of the signal, the ideal amplitude-frequency response of the Farrow filter is constant and the phase-frequency response is linear phase within the range of the digital angular frequency omega epsilon [0.35 pi, 0.65 pi ]. The mean square error of amplitude-frequency response in the bandwidth is required to be better than 0.1dB, and the mean square error of delay quantity is required to be better than 10 ps.
Fig. 3a) and b) respectively show the relationship between the amplitude-frequency response mean square error and the delay amount mean square error of the variable fractional delay Farrow filter and the group number and the order thereof. On the premise of meeting the performance requirement, the resource consumption is reduced as much as possible, 7 groups of Farrow filters with 6 orders are selected, and the structural block diagram is shown in fig. 4. The FIR filters 228-221 forming the Farrow filter are arranged in the sequence of FIR 7-FIR 0, and the coefficient of each FIR filter is configured with the calculated coefficient, which is listed in table 1, and only needs to be configured once and called by IP core. The Farrow filters of the N channels have the same structure and coefficient, and the difference between the Farrow filters is that the delay control parameter p260 is different.
TABLE 17 set of 6 th order Farrow filter coefficients
Step 2, setting the coefficient of the variable FIR filter of the digital low-pass filtering and channel equalization integrated module 113 as the coefficient of a 42-order digital low-pass FIR filter;
fig. 5 shows an implementation scheme and interconnection relationship of the digital down-conversion module 112 and the digital low-pass filtering and channel equalization integrated module 113.
According to the center frequency and the sampling frequency of the signal, the values of the orthogonal local oscillation sequence 321 and the local oscillation sequence 322 listed in table 2 are calculated. The processing of quadrature digital down-conversion 320 is to periodically take the original value, zero, and inverse of the original data.
TABLE 2 Quadrature local oscillator sequence values
Orthogonal local oscillator sequence
|
n=0
|
n=1
|
n=2
|
n=3
|
n=4
|
n=5
|
n=6
|
n=7
|
…
|
cos(w0n)
|
1
|
0
|
-1
|
0
|
1
|
0
|
-1
|
0
|
…
|
sin(w0n)
|
0
|
1
|
0
|
-1
|
0
|
1
|
0
|
-1
|
… |
Digital low pass filtering and channel equalization 330 performs 4 times filtering decimation and compensates for mismatch between channels. The HBF331 is a half-band filter, realizes 2 times of extraction, and meets the requirements of 15MHz passband cut-off frequency, 85MHz stopband cut-off frequency and-100 dB stopband attenuation. FIR332 is a digital low-pass FIR filter, which realizes 2 times of extraction, the cut-off frequency of the pass band is 15MHz, the cut-off frequency of the stop band is 25MHz, the pass band ripple is 0.1dB, the stop band attenuation is-110 dB, and the order of the filter is 42; equalizer 333 implements a channel equalization function.
And 3, setting the integral multiple of the clock period and the fractional multiple delay of the variable integer/fractional delay module 111 to 0.
The channel equalization mode is executed by the following steps:
step A, the digital low-pass filtering and channel equalizer coefficient calculating module 130 calculates the equalizer amplitude-phase characteristics of each channel according to the output data of each channel digital down-conversion module 112 by using a frequency domain channel equalization algorithm (obtained through Adaptive channel equalization for space-time Adaptive processing);
step B, the digital low-pass filtering and channel equalizer coefficient calculating module 130 synthesizes the amplitude-phase characteristics of the 42-order digital low-pass FIR filter 332 and the equalizer 333 of each channel according to the formula (3) to obtain the joint amplitude-phase characteristics of each channel, and optimizes the 42-order digital low-pass FIR filter coefficient of each channel to approach the respective joint amplitude-phase characteristics;
and C, setting the coefficient of the variable coefficient FIR filter 1131 of each channel digital low-pass filtering and channel equalization integrated module 113 to be a corresponding 42-order optimization coefficient.
The implementation scheme of the digital low-pass FIR filter 332 and the equalizer 333 after synthesis optimization is shown in fig. 6, where complex filtering is converted into real filtering, and 401 and 402 are inputs of in-phase data and quadrature data, respectively; they enter the real part coefficient 403 and imaginary part coefficient 404 of the variable coefficient FIR filter, respectively, and then add and subtract to obtain the in-phase data output 405 and the quadrature data output 406.
The execution steps of the normal working mode are as follows:
step a, calculating delay values of all channels according to the requirement of change of beam pointing direction and the formula (4), converting the delay values into integral multiples of clock periods and fractional time delays, and adjusting in real time;
τn=nd sinθ0/c n=1,2,...,N (4)
wherein d represents the array element spacing, θ0Denotes the beam pointing direction, c denotes the speed of the electromagnetic wave propagating in the vacuum, N denotes the number of channels, τnIndicating the delay value for each channel.
Step b, setting the integral multiple of the clock period and the fractional multiple delay of each channel variable integer/fractional delay module 111 according to the result;
step c, all modules of the whole circuit work by adopting a parallel pipeline structure, namely, data is delayed by a corresponding clock period through the variable integer/fraction delay module 111; the delayed data enters the digital down-conversion module 112 again to become the I/Q data of the baseband quadrature; the I/Q data with orthogonal baseband is processed by a digital low-pass filtering and channel equalization integrated module 113 to obtain broadband beam forming data with low data rate; the wideband beamformed data enters the data synthesis module 120 and is output as a beam.
The invention can be widely applied to radar digital array antenna, communication intelligent antenna, microphone array, sonar array and other systems.